JP7256935B2 - Dictionary creation device and dictionary creation method - Google Patents

Description

The present invention relates to a dictionary creation device and a dictionary creation method.

2. Description of the Related Art Techniques for analyzing logs output from a device are known in order to appropriately respond to inquiries from customers who have introduced the device when performing support work for the device. In recent years, machine learning may be used in such techniques. In machine learning, for example, characteristic phrases (character strings) are extracted from logs and an inference model is created to infer answers from the extracted results. Some of these characteristic phrases are registered in dictionaries, and it is known that whether or not the dictionaries are appropriately created greatly affects the inference accuracy of the inference model.

JP 2016-134124 A JP-A-11-154160

Phrases to be registered in the dictionary are important phrases that characterize the logs of each device, and are picked up in advance by the person in charge. .

Therefore, when a device to be newly supported is added, it may take a lot of man-hours and time to create a dictionary for the device. In addition, there are cases in which a phrase to be registered in the dictionary is not known until after a certain amount of support work has been performed.

In one aspect, an object of the present invention is to provide a dictionary creation device and a dictionary creation method capable of easily creating a dictionary for registering character strings to be extracted from device logs.

In one aspect, the dictionary creation device extracts a plurality of character strings having a common part from a first dictionary that registers character strings extracted from a log of the first device, and extracts the extracted plurality of character strings. a first identifying unit that identifies the type of the character string of the first device, a second identifying unit that extracts the character string including the common part from the log of the first device and identifies the type of the extracted character string; 2 a first determination unit that determines whether the type of the character string specified by the first specification unit is limited with respect to the type of the character string specified by the specification unit; If it is limited, a character string partly including the common part is extracted from the log of a second device different from the first device, and the type of the extracted character string is specified and identified a second judging unit for judging whether the degree of matching between the type of the character string and the type of the character string identified by the second identifying unit is equal to or greater than a predetermined degree; a dictionary editing unit that includes the plurality of character strings extracted by the unit in a second dictionary that registers character strings extracted from the log of the second device.

A dictionary for registering character strings to be extracted from device logs can be easily created.

It is a figure which shows roughly the usage scene of the dictionary creation apparatus which concerns on one Embodiment. It is a figure which shows an example of the analysis method of the log by a support operation assistance apparatus. It is a figure which shows the hardware constitutions of a dictionary creation apparatus. It is a functional block diagram of a dictionary creation device. It is a figure which shows the outline|summary of the process of a dictionary creation apparatus. It is a flowchart which shows the whole process of a dictionary creation apparatus. FIG. 7 is a flowchart showing detailed processing in step S10 of FIG. 6. FIG. FIG. 7 is a flowchart showing detailed processing in step S14 of FIG. 6. FIG. FIG. 9(a) is a diagram showing an example of a dictionary of an existing device, and FIG. 9(b) is a diagram showing a processing target. 10(a) to 10(d) are diagrams for explaining the process of step S10 in FIG. FIG. 11(a) is a diagram showing an example of the log of the existing device, and FIG. 11(b) is a diagram showing phrases extracted from the log of the existing device. FIG. 10 is a diagram for explaining a state in which variable parts specified in a dictionary of an existing device are narrowed down with respect to variable parts specified in a log of an existing device; FIG. 10 is a diagram for explaining a state in which variable parts specified in a dictionary of an existing device are not narrowed down with respect to variable parts specified in a log of an existing device; FIG. 14(a) is a diagram showing another example in which the variable part specified in the dictionary of the existing device is not narrowed down with respect to the variable part specified in the log of the existing device, and FIG. ) is a diagram showing another example of the narrowed down state. FIGS. 15(a) to 15(d) are diagrams showing examples when the degree of matching is equal to or higher than the threshold. 16(a) and 16(b) are diagrams showing examples in which the degree of matching is less than the threshold. FIG. 10 is a diagram showing phrases registered in the dictionary of the new device in one case;

Hereinafter, a dictionary creation device according to one embodiment will be described in detail with reference to FIGS. 1 to 16. FIG.

FIG. 1 schematically shows a usage scene of the dictionary creation device 10 of this embodiment. As shown in FIG. 1, in this embodiment, when the person in charge of the support center receives an inquiry from a customer regarding a device (for example, a network device) that has been introduced to the customer, the person in charge of the support center take advantage of A dictionary creation device 10 is connected to this support work support device.

When a problem or the like occurs in a network device, the customer sends an inquiry about the problem by sending a log output from the network device to the support center. The person in charge of the support center inputs the received log into the support operation support device, analyzes it, and refers to the response pattern obtained to respond to the customer. Here, the dictionary creation device 10 acquires in advance the log output from the network device, creates and provides a dictionary to be used for analysis by the support work support device.

FIG. 2 is a diagram illustrating an example of a log analysis method by the support work support device. In this embodiment, the support work support device uses machine learning to analyze the log and output the response pattern. In machine learning, as preprocessing, many phrases (character strings) are extracted from past inquiry logs, and vector values are calculated for each phrase (Word_1, Word_2, . . . Word_n) in each log. This vector value is, for example, a count value of the number of occurrences of each phrase included in the inquiry log. At this time, the phrases extracted from the inquiry log are important phrases that characterize the log of each network device, and some of the forcedly extracted phrases are stored in the dictionary created by the dictionary creation device 10. . Phrases stored in the dictionary are, for example, phrases contained only in logs corresponding to certain answer patterns. Then, as shown in Table A of FIG. 2, the vector value of each phrase in each inquiry log is associated with the answer pattern (label) number, and based on this, an inference model is created.

On the other hand, when a new inquiry log is entered by a support center representative, as with past inquiry logs, as preprocessing, many phrases are extracted from the new inquiry log, vector values are calculated, and the calculated vector values are inferred. Enter the model. As a result, answer pattern candidates corresponding to the new inquiry log are output together with the confidence value (see Table B in FIG. 2).

By referring to Table B, the person in charge of the support center can select an appropriate response pattern and respond to the customer's inquiry.

By the way, when a new network device (new device) for which an inference model has not been created is to be supported, it is necessary to create a new dictionary for the new device. If the person in charge performs this work, it may require a great deal of man-hours and time. Therefore, in the present embodiment, the dictionary creation device 10 uses a dictionary (first dictionary) used to support existing devices (devices that are already supported) and a dictionary to be used to support new devices. (second dictionary).

As the existing device when creating a dictionary for the new device, a device similar to the new device, such as being of the same vendor as the new device, can be adopted. This may allow the dictionary of the new device to be created more accurately. However, the present invention is not limited to this, and various existing devices may be adopted as the existing device regardless of the new device.

The dictionary creation device 10 will be described in detail below. FIG. 3 shows the hardware configuration of the dictionary creation device 10. As shown in FIG. The dictionary creation device 10 includes a CPU (Central Processing Unit) 90, a ROM (Read Only Memory) 92, a RAM (Random Access Memory) 94, a storage unit (here, a HDD (Hard Disk Drive)) 96, a network interface 97, a display unit 93, an input unit 95, a portable storage medium drive 99, and the like. Each component of the dictionary creation device 10 is connected to the bus 98 . In the dictionary creation device 10, the CPU 90 executes a program (including a dictionary creation program) stored in the ROM 92 or HDD 96, or a program read from the portable storage medium 91 by the portable storage medium drive 99. 4 are realized. 4 may be implemented by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

FIG. 4 shows a functional block diagram of the dictionary creation device 10. As shown in FIG. In the dictionary creation device 10, the CPU 90 executes a program so that the first extracting section 20, the second extracting section 22, the narrowing determination section 24 as the first determining section, the third extracting section 26, and the degree of matching determination section 28 , and a registered phrase determining unit 30 as a dictionary editing unit. The function of each unit of the dictionary creation device 10 will be described below with reference to FIG. 5 showing an overview of the processing of the dictionary creation device 10.

The first extraction unit 20 extracts and groups partially common phrases from the dictionary of the existing device (step S100 in FIG. 5), and extracts common parts (called fixed parts) of the extracted phrases and uncommon parts. (referred to as a variable portion) (step S110). Hereinafter, this process of dividing into the fixed part and the variable part will be referred to as "shaping". Although FIG. 5 shows only a group of phrases having a fixed part (HARDWARE FAULT), the first extraction unit 20 extracts all groups of phrases having a common part from the dictionary of the existing device. and shape it.

The second extraction unit 22 extracts phrases including fixed parts of the phrases extracted by the first extraction unit 20 from the log of the existing device, groups them, and shapes them (step S120). Note that FIG. 5 shows only a group of phrases having a fixed portion (HARDWARE FAULT).

The narrowing-down determination unit 24 identifies the type (variation) of the variable portion corresponding to the fixed portion (for example, “HARDWARE FAULT”) extracted by the second extraction unit 22 . In addition, the narrowing-down determination unit 24 identifies the type (variation) of the variable portion corresponding to a certain fixed portion (for example, “HARDWARE FAULT”) extracted by the first extraction unit 20 . Then, by referring to the type of each variable part, it is determined whether the type of the variable part specified in the dictionary of the existing device is narrowed down to the type of the variable part specified in the log of the existing device. It judges (step S130). That is, in the example of FIG. 5, the narrowing-down determination unit 24 determines that the narrowed-down state is in the case where the variable part indicated by symbol F2 includes all of the variable parts indicated by symbol F1 and does not completely match. to decide. Here, the fact that phrases are narrowed down in the dictionary of the existing device means that there is a high possibility that only characteristic phrases among the phrases existing in the log are registered in the dictionary.

The third extraction unit 26 extracts a phrase including a fixed part (for example, "HARDWARE FAULT") from the log of the new device from the log of the existing device when the narrowing-down determination unit 24 determines that the state is narrowed down. It is extracted and shaped (step S140).

The matching degree determination unit 28 determines the degree of matching (matching ratio ) is determined (step S150). Details of the degree of matching will be described later.

The registration phrase determination unit 30 determines phrases to be registered in the dictionary of the new device based on the determination result of the narrowing down determination unit 24 and the determination result of the matching degree determination unit 28 (step S160). Details of the processing of the registered phrase determination unit 30 will be described later.

(Regarding the processing of the dictionary creation device 10)
Hereinafter, the processing of the dictionary creation device 10 will be described in detail along the flow charts of FIGS. 6 to 8 and with appropriate reference to other drawings. 6 is a flow chart showing the overall processing of the dictionary creation device 10, FIG. 7 is a flow chart showing the detailed processing of step S10 in FIG. 6, and FIG. 8 is a flow chart showing the detailed processing of step S14 in FIG. is.

In the process of FIG. 6, first, in step S10, the first extraction unit 20 extracts a phrase from the dictionary of the existing device and executes processing to shape it. Specifically, in this step S10, the process according to the flowchart of FIG. 7 is executed.

(Detailed processing of step S10)
In the process of FIG. 7, first, in step S40, the first extraction unit 20 sets a base point and a comparison target. For example, assume that the dictionary of the existing device has 11 phrases as shown in FIG. 9(a). In this case, as shown in FIG. 9(b), the first extraction unit 20 processes the entire dictionary of the existing device, sets the first phrase (the first phrase) as a base point, and sets the remaining second to The 11th phrase is set as a comparison target.

Next, in step S42, the first extraction unit 20 compares the base phrase and each comparison target phrase. In this case, the first extraction unit 20 distinguishes, on a word-by-word basis, words that match the words contained in the base phrase and words that do not match.

Then, in step S44, the first extraction unit 20 groups comparison objects having the same pattern of the position of the matching word and the position of the non-matching word, and the base point. For example, in the example of FIG. 9B, the reference point and the comparison objects 1 to 6 are the matching words (HARDWARE FAULT) in the first and second words, and the non-matching word in the third word. The pattern of word positions is the same. Therefore, the first extraction unit 20 groups these phrases into one group.

Next, in step S46, the first extraction unit 20 shapes the phrase by setting the matching word (HARDWARE FAULT) of each phrase included in the group as a fixed part and the non-matching word (other than HARDWARE FAULT) as a variable part. FIG. 10(a) shows a state (extracted phrase (after shaping)) in which the base point and comparison objects 1 to 6 are shaped.

Next, in step S48, the first extraction unit 20 excludes the grouped items from the processing targets. In this case, the object to be processed is as shown in FIG. 10(b), which is obtained by removing the extracted phrases of FIG. 10(a) from the dictionary of the existing device shown in FIG. 9(a).

Next, in step S50, the first extraction unit 20 determines whether or not there are any remaining processing targets. If the determination in step S50 is affirmative, the process returns to step S40. After that, the first extraction unit 20 sets a base point and comparison targets 1, 2, and 3 for the processing target in FIG. 10B (S40), and groups the base point and comparison targets 1 and 2 (S44 ) and shaping (S46). As a result, an extracted phrase (after shaping) as shown in FIG. 10(c) is obtained.

After that, the grouped phrases are excluded from the processing targets, but since "Service Level * has occurred with core" remains in the processing targets, the determination in step S50 is affirmative, and the process returns to step S40. After that, by further executing steps S40 to S50, an extracted phrase (after shaping) as shown in FIG. 10(d) is obtained. "Service Level * has occurred with core" includes "*" indicating a wildcard (arbitrary character string). In such a case, the wild card (*) is treated as a non-matching word (variable part), and the rest is treated as a matching word (fixed part). Thereafter, if the determination in step S50 is negative, the process proceeds to step S12 in FIG.

Returning to FIG. 6, when the process proceeds to step S12, the narrowing down determination unit 24 selects one fixed part from the group (extracted phrase (after shaping)) obtained in step S10. Here, it is assumed that the narrow-down determination unit 24 selects the fixed part "HARDWARE FAULT" from among FIGS. 10(a), 10(c), and 10(d).

Next, in step S14, the narrowing determination unit 24 and the second extraction unit 22 perform processing for confirming whether the selected fixed portion is narrowed down or not narrowed down. Specifically, in this step S14, the process according to the flowchart of FIG. 8 is executed.

(Detailed processing of step S14)
In the process of FIG. 8, first, in step S60, it is determined whether or not the variable part corresponding to the selected fixed part is a wild card. If the determination in step S60 is negative, the process proceeds to step S61.

In step S61, the narrowing-down determination unit 24 identifies the type (variation) of the variable part corresponding to the selected fixed part (HARDWARE FAULT) in the existing device dictionary (FIG. 9A). Here, the narrowing-down determination unit 24 identifies seven types ((I/OMod-EX), (MDirAgnt 100), . . . , (PRB)) shown in FIG. 10(a). It can be said that the seven types specified here are the number of types of phrases (FIG. 10(a)) extracted from the dictionary of the existing device.

Next, in step S62, the second extractor 22 extracts a phrase corresponding to the selected fixed part (HARDWARE FAULT) in the log of the existing device. Further, the narrowing-down determination unit 24 specifies the type (variation) of the variable portion. More specifically, the second extractor 22 prepares the log of the existing device as shown in FIG. 11(a). Note that the log of the existing device is a log sent when an inquiry is received from the customer. Then, the second extraction unit 22 extracts phrases including the selected fixed part (HARDWARE FAULT) from the log of the existing device, and divides them into the fixed part and the variable part (shaping) as shown in FIG. do). In the example of FIG. 11B, the narrowing-down determination unit 24 specifies eight types ((I/OMod-EX), (MDirAgnt 100), . It can be said that the eight types specified here are the number of types of phrases (FIG. 11(b)) extracted from the log of the existing device.

Next, in step S64, the narrowing-down determination unit 24 determines the type of the variable part specified in the dictionary of the existing device (FIG. 10A) and the type of the variable part specified in the log of the existing device (FIG. 11B )) and compare. The comparison checks to see if the type of variable portion identified in the dictionary of the existing device is specific to the type of variable portion identified in the log of the existing device. Here, "restrictive" means that each variable part specified in the dictionary of the existing device is included in the variable part specified in the log of the existing device and is not a perfect match.

Next, in step S66, the narrowing-down determination unit 24 determines whether or not it is restrictive. In the examples of FIGS. 10A and 11B, as shown in FIG. 12, the variable parts (8 types) specified in the dictionary of the existing device are added to the variable parts (7 types) specified in the log of the existing device. Type) and other variable parts are included, so it is judged to be limited. In this case, the determination in step S66 of FIG. 8 is affirmative, and the process proceeds to step S68.

When the process proceeds to step S68, the narrowing down determination unit 24 determines that the narrowing down has been performed, ends all the processes in FIG. 8, and proceeds to step S16 in FIG.

On the other hand, if the determination in step S66 is negative, the process proceeds to step S70. For example, as shown in the lower part of FIG. 13, if there are seven types of variable parts specified in the log of the existing device, as shown in the upper part of FIG. It is assumed that there are 7 types and that they are completely matched. In this case, since it is not restrictive, the determination in step S66 is negative, the process proceeds to step S70, and the narrowing down determination unit 24 determines that the narrowing down is not performed. After that, the process proceeds to step S16 in FIG.

For example, as shown in the upper left diagram of FIG. 14(a), it is assumed that "Service Level, has occurred with core" is selected as the fixed part and two types of phrases are extracted from the dictionary of the existing device. It is also assumed that two types of phrases shown in the lower right diagram of FIG. 14A are extracted from the log of the existing device as phrases containing the selected fixed part "Service Level, has occurred with core". In such a case, it is determined that the dictionaries of the existing devices have not been narrowed down because the types of the variable portions are exactly the same in the upper left diagram and the lower right diagram of FIG. 14(a) (S70). On the other hand, as shown in the upper left diagram of FIG. 14B, it is assumed that "Service Level, has occurred with core" is selected as the fixed part and one type of phrase is extracted from the dictionary of the existing device. It is also assumed that two types of phrases shown in the lower right diagram of FIG. 14B are extracted from the log of the existing device as phrases containing the selected fixed part "Service Level, has occurred with core". In such a case, it is determined that the dictionaries of existing devices have been narrowed down (S68).

By the way, it is assumed that the fixed part "Service Level, has occurred with core" in FIG. 10D is selected as the fixed part in step S12 of FIG. In this case, since the variable part is a wild card, the determination in step S60 is affirmative in the process of FIG. In step S70, the narrowing down determination unit 24 determines that the dictionaries of the existing devices have not been narrowed down.

Returning to FIG. 6, when the process moves to step S16, the narrowing-down determination unit 24 determines whether or not the narrowing-down is performed. If the determination in step S16 is affirmative, the process proceeds to step S18.

When the process proceeds to step S18, the third extraction unit 26 extracts the phrase corresponding to the selected fixed part (HARDWARE FAULT) from the log of the new device, and separates it into the fixed part and the variable part for shaping. In this case, the third extraction unit 26 searches for phrases including "HARDWARE FAULT" in the log of the new device. For example, assume that the phrases extracted from the log of the new device are the phrases shown in FIG. 15(a).

Next, in step S20, the degree-of-match determination unit 28 determines whether or not a phrase including the selected fixed part has been extracted from the log of the new device. If at least one phrase is extracted as shown in FIG. 15A, the determination in step S20 is affirmative, and the process proceeds to step S22.

In step S22, the degree-of-match determination unit 28 determines the phrase extracted from the log of the new device (FIG. 15(a)), the phrase extracted from the log of the existing device (see FIG. 15(b)), Calculate the degree of matching. Specifically, the matching degree determination unit 28 uses the phrase extracted from the log of the existing device (FIG. 15B) and the phrase extracted from the log of the new device (FIG. 15A) to Create a table as shown in FIG. 15(c). Here, in the table of FIG. 15(c), all phrases (fixed parts and variable parts) of FIGS. Whether or not it is extracted from the log of the device is indicated by "yes"/"no". From the table of FIG. 15(c), the degree-of-match determination unit 28 obtains the percentage of phrases extracted from both the log of the existing device and the log of the new device in all phrases as the "degree of match." In the example of FIG. 15(c), the ratio of the number of phrases for which "Yes" is stored in the log column of the existing device and "Yes" is stored in the log column of the new device to the total number of extracted phrases. (Number of matches/number of all phrases=6/9≈0.67) is obtained as the degree of matching.

Next, in step S24, the degree-of-match determination unit 28 determines whether or not the degree of match is greater than or equal to a threshold. For example, assuming that the threshold is 0.6, in the example of FIG. 15(c), the determination in step S24 is affirmative, and the process proceeds to step S26. Incidentally, in step S24, it can be said that it is determined whether or not the narrowing down of the existing devices is also effective in the new device.

If the process proceeds to step S26, it can be assumed that the variation of the variable part specified in the log of the existing device is substantially the same as the variation of the variable part specified in the log of the new device. In this case, it is considered effective to use the same narrowing rule as the existing device for the new device. Therefore, in step S26, the registration phrase determination unit 30 registers the phrase having the selected fixed part (HARDWARE FAULT) in the dictionary of the existing device as it is in the dictionary of the new device (Fig. 15(d) )reference). After that, the process proceeds to step S32.

On the other hand, FIG. 16A shows an example in which the degree of matching is less than the threshold. In the example of FIG. 16A, the presence/absence of phrases included in the log of the new device and the presence/absence of phrases included in the log of the existing device do not match very much. In FIG. 16(a), when the degree of matching is calculated in the same manner as in FIG. 15(c), the degree of matching is 4/11≈0.36, and the threshold value is less than 0.6. In such a case, the determination in step S24 is negative, and the process proceeds to step S30. If the process proceeds to step S30, it can be assumed that the variation of the variable part specified in the log of the existing device is significantly different from the variation of the variable part specified in the log of the new device, so the same narrowing rule as that of the existing device should be used. can't. Therefore, in step S30, the registered phrase determination unit 30 designates the variable part as a wild card (*) (arbitrary character string) and registers it in the dictionary of the new device as shown in FIG. 16(b). That is, information is registered in the dictionary of the new device to extract all phrases having a fixed part (HARDWARE FAULT). After that, the process proceeds to step S32.

By the way, when the determination in step S16 is negative, that is, when the phrases extracted from the dictionary of the existing device have not been narrowed down to the phrases extracted from the log of the existing device, the registered phrase determining unit 30 moves to step S30. In this case as well, the registered phrase determination unit 30 designates the variable part as a wildcard and registers it in the dictionary of the new device as shown in FIG. 16(b). It should be noted that the case where the determination in step S16 is negative also includes the case where the variable part of the phrase extracted from the dictionary of the existing device is a wild card, as shown in FIG. 10(d). In this case, the registered phrase determination unit 30 registers the phrase itself extracted from the dictionary of the existing device in the dictionary of the new device in step S30, as shown in FIG. 16(c).

If the determination in step S20 in the process of FIG. 6 is negative, that is, if no phrase containing the selected fixed part (HARDWARE FAULT) is extracted from the log of the new device, the registered phrase determination unit 30 moves to step S28. When proceeding to step S28, the registered phrase determining unit 30 does not register the phrase having the selected fixed part (HARDWARE FAULT) in the dictionary of the new device. After that, the process proceeds to step S32.

After moving to step S32, the narrowing-down determination unit 24 determines whether or not all the fixed portions have been selected. If the determination in step S32 is negative, the process returns to step S12, and the processes of steps S12 to S32 are repeatedly executed until the determination in step S32 is positive. Then, when the determination in step S32 is affirmative, all the processing in FIG. 6 ends.

As can be seen from the above description, in this embodiment, the first extraction unit 20 and the narrowing-down judgment unit 24 extract phrases from the dictionary of the existing device, A function as a first specifying unit for specifying is realized. In addition, the second extraction unit 22 and the narrowing determination unit 24 realize a function as a second identification unit that extracts phrases from the log of the existing device and identifies the type of the extracted phrase (type of variable part). there is Furthermore, the third extraction unit 26 and the degree of matching determination unit 28 extract phrases from the log of the new device, identify the type of the extracted phrase (type of variable part), and function as a second determination unit that determines whether the degree of matching with the type of phrase (type of variable part) extracted from the log of 1 is above a predetermined level.

As described in detail above, according to the present embodiment, the first extraction unit 20 extracts a plurality of partially common phrases from a dictionary (existing device dictionary) that registers phrases to be extracted from logs of an existing device. (S10), and the narrowing-down determination unit 24 identifies the type of variable portion other than the common portion (fixed portion) (S61). Further, the second extraction unit 22 extracts phrases including fixed parts from the log of the existing device, and the narrowing-down determination unit 24 specifies the types of variable parts included in the extracted phrases (S62). In addition, the narrowing-down determination unit 24 determines whether the type of the variable part specified from the dictionary of the existing device is limited to the type of the variable part specified from the log of the existing device (S66 to S70, S16). As a result of this determination, if it is limited (S16: affirmative), the third extraction unit 26 extracts phrases including fixed parts from the log of the new device (S18). Also, the matching degree determination unit 28 determines whether the degree of matching between the types of variable portions included in the phrase extracted from the log of the new device and the type of variable portions specified from the log of the existing device is equal to or greater than a threshold value ( S24). Then, if the number of phrases is equal to or greater than the threshold (S24: Yes), the registered phrase determination unit 30 registers the phrases extracted from the dictionary of the existing device as they are in the dictionary of the new device (S26). As a result, in the present embodiment, when characteristic phrases are narrowed down and registered in the dictionary of the existing device, and narrowing down of characteristic phrases is effective in the log of the new device, the dictionary of the existing device characteristic phrases can be automatically registered in the new device's dictionary. Therefore, it is possible to register appropriate phrases in the dictionary of the new device without spending work man-hours.

In addition, in this embodiment, in addition to the case where the new device is a device of the series of the existing device, when the configuration of the new device (line card, interface card, chip, parts) is common or similar to the existing device , it is possible to automatically create a new device dictionary without spending a lot of work.

Here, when creating a dictionary for a new device, it is conceivable to use all the dictionaries of existing devices. However, in this case, even when the degree of matching is less than the threshold (case of FIG. 16A), only the narrowed down phrases are registered in the dictionary of the new device. Therefore, since machine learning using appropriate phrases is no longer performed, there is a risk that the inference accuracy of the answer pattern will decrease. On the other hand, in the present embodiment, all phrases including fixed parts (phrases whose variable parts are designated as wildcards) are registered in the dictionary of the new device when the degree of matching is not equal to or greater than the threshold. As a result, appropriate phrases (characteristic phrases in the log of the new device but not in the log of the existing device) are extracted from the log of the new device. Therefore, it is possible to improve the accuracy of answer pattern inference as compared with the case where the dictionary of the existing device is registered as it is in the dictionary of the new device.

On the other hand, if the match rate is low, a method of not registering the phrase containing the selected fixed part in the dictionary of the new device is also conceivable. However, as long as the phrase containing the selected fixed part is included in the log of the new device, there is a high possibility that at least the selected fixed part is an important phrase that characterizes the log of the new device. Therefore, in the present embodiment, such phrases are registered in the dictionary of the new device with a wildcard specification, and are forcibly extracted. By doing so, it is possible to improve the inference accuracy at the time of starting support for the new device.

Further, in the present embodiment, the narrowing-down determination unit 24 includes all types of variable portions specified in the dictionary of the existing device in the types of variable portions specified in the log of the existing device, and , and if they do not completely match, it is determined to be limited (S14, FIG. 8). Thereby, it is possible to appropriately determine whether or not a characteristic phrase is registered in the dictionary of the existing device.

Further, in the present embodiment, if the result of determination by the narrowing-down determination unit 24 is that it is not restrictive (S16: No), the registered phrase determination unit 30 selects all phrases including the selected fixed part (the variable part is wild). card-specified phrase) is registered in the dictionary of the new device (S30). As a result, it is possible to suppress omission of extraction of characteristic phrases from the log of the new device.

Further, in the present embodiment, when a phrase containing the selected fixed part cannot be extracted from the log of the new device (S20: No), the registered phrase determination unit 30 adds the phrase containing the selected fixed part to the new device. It is not registered in the dictionary of the device (S28). This makes it possible to avoid registering phrases that are highly likely to be unnecessary when obtaining answer patterns from the log of the new device in the dictionary of the new device.

In addition, in the present embodiment, when a phrase including a wildcard designation is registered in the dictionary of the existing device, the registered phrase determination unit 30 includes the phrase in the dictionary of the new device (S60: affirmative, S70, S16 : No, S30). As a result, it is possible to suppress omission of extraction of characteristic phrases from the log of the new device.

Further, in the present embodiment, types of variable parts other than fixed parts are specified in phrases, and whether or not they are narrowed down is determined and the degree of matching is calculated. Therefore, the amount of processing can be reduced compared to the case of determining whether or not the phrases have been narrowed down and calculating the degree of matching using the entire phrase.

In the above embodiment, phrases extracted from dictionaries and logs are divided into fixed parts and variable parts and shaped, and the types of the variable parts are used to determine whether or not the phrases have been narrowed down and to calculate the degree of matching. has been described, but it is not limited to this. That is, it is possible to determine whether or not the phrases have been narrowed down and calculate the degree of matching using the types of phrases extracted from a dictionary or log.

Note that the processing functions described above can be realized by a computer. In that case, a program is provided that describes the processing contents of the functions that the processing device should have. By executing the program on a computer, the above processing functions are realized on the computer. A program describing the processing content can be recorded in a computer-readable storage medium (excluding carrier waves).

When a program is distributed, it is sold in the form of a portable storage medium such as a DVD (Digital Versatile Disc) or a CD-ROM (Compact Disc Read Only Memory) on which the program is recorded. It is also possible to store the program in the storage device of the server computer and transfer the program from the server computer to another computer via the network.

A computer that executes a program stores, for example, a program recorded on a portable storage medium or a program transferred from a server computer in its own storage device. The computer then reads the program from its own storage device and executes processing according to the program. The computer can also read the program directly from the portable storage medium and execute processing according to the program. In addition, the computer can also execute processing in accordance with the received program each time the program is transferred from the server computer.

The embodiments described above are examples of preferred implementations of the present invention. However, the present invention is not limited to this, and various modifications can be made without departing from the spirit of the present invention.

In addition, the following additional remarks will be disclosed with respect to the above description of the embodiment.
(Appendix 1) A plurality of character strings having a common part are extracted from a first dictionary for registering character strings extracted from the log of the first device, and the types of the extracted character strings are determined. a first identifying part to identify;
a second identification unit that extracts a character string including the common part from the log of the first device and identifies the type of the extracted character string;
a first determining unit that determines whether the type of character string specified by the first specifying unit is limited with respect to the type of character string specified by the second specifying unit;
If the result of the determination by the first determination unit is limited, a character string including the common part is extracted from the log of the second device different from the first device, and the extracted character a second determining unit that identifies the type of the string and determines whether the degree of matching between the identified character string type and the character string type identified by the second identifying unit is equal to or greater than a predetermined value;
a dictionary editing unit including the plurality of character strings extracted by the first identifying unit in a second dictionary for registering character strings extracted from the log of the second device when the degree of matching is equal to or greater than a predetermined value; and,
A dictionary creation device comprising:
(Supplementary Note 2) The first determining unit determines that the types of character strings specified by the second specifying unit include all of the types of character strings specified by the first specifying unit, and 1. The method according to appendix 1, wherein if the type of the character string specified by the specifying unit and the type of the character string specified by the first specifying unit do not completely match, it is determined that the character string is limited. Dictionary creation device.
(Supplementary note 3) If the result of the determination by the first determination unit is that the dictionary editing unit is not restrictive, the dictionary editing unit provides information for extracting all character strings having the common part from the log of the second device. , is included in the second dictionary.
(Supplementary Note 4) The dictionary editing unit stores information for extracting all character strings including the common part from the log of the second device in the second dictionary when the degree of matching is not equal to or greater than a predetermined value. 4. The dictionary creation device according to any one of appendices 1 to 3, characterized in that it is included in.
(Additional Note 5) When the second determination unit fails to extract the character string including the common portion from the log of the second device, the dictionary editing unit may extract the character string including the common portion from the log of the second device. 5. The dictionary creation device according to any one of Appendices 1 to 4, characterized in that it does not include in the dictionary of.
(Supplementary Note 6) The dictionary editing unit includes the character string in the second dictionary when the character string including the part where the arbitrary character is specified is registered in the first dictionary. The dictionary creation device according to any one of appendices 1 to 5.
(Appendix 7) The dictionary according to any one of Appendices 1 to 6, characterized in that the process of identifying the type of the character string is a process of identifying the type of a portion of the character string other than a common portion. creation device.
(Appendix 8) A plurality of character strings having a common part are extracted from the first dictionary that registers the character strings extracted from the log of the first device, and the types of the extracted character strings are determined. identify,
Extracting a character string containing the common part from the log of the first device, identifying the type of the extracted character string,
determining whether the type of character string identified from the first dictionary is limited to the type of character string identified from the log of the first device;
As a result of the determination, if it is limited, a character string partly including the common part is extracted from the log of the second device different from the first device, and the type of the extracted character string is specified. and determining whether the degree of matching between the type of the specified character string and the type of the character string specified from the log of the first device is equal to or greater than a predetermined value;
If the degree of matching is equal to or greater than a predetermined value, include the plurality of character strings extracted from the first dictionary in a second dictionary for registering character strings extracted from the log of the second device;
A dictionary creation method characterized in that the processing is executed by a computer.
(Supplementary Note 9) In the processing for determining whether it is restrictive, all the types of character strings specified from the dictionary of the first device are included in the types of character strings specified from the log of the first device. is included and the type of character string identified from the log of the first device and the type of character string identified from the dictionary of the first device do not completely match, it is determined to be limited The dictionary creation method according to Supplementary Note 8, characterized by:
(Supplementary Note 10) If the result of the processing for determining whether it is restrictive is not restrictive, the information for extracting all character strings having the common part from the log of the second device is added to the log of the second device. 10. The dictionary creation method according to appendix 8 or 9, wherein the dictionary is included in the dictionary of appendix 8 or 9.
(Appendix 11) The second dictionary includes information for extracting all character strings including the common portion from the log of the second device when the degree of matching is not equal to or higher than a predetermined value. The dictionary creation method according to any one of Supplements 8 to 10.
(Supplementary note 12) Supplementary note 8 characterized by not including the character string including the common part in the second dictionary when the character string including the common part cannot be extracted from the log of the second device. 12. The method for creating a dictionary according to any one of 11.
(Supplementary Note 13) Supplementary notes 8 to 12 characterized in that, when a character string including a portion where an arbitrary character is specified is registered in the first dictionary, the character string is included in the second dictionary. A method for creating a dictionary according to any one of the above.
(Appendix 14) The dictionary according to any one of Appendices 8 to 13, characterized in that the process of identifying the type of the character string is a process of identifying the type of a portion of the character string other than a common portion. How to make.

10 dictionary creation device 20 first extraction unit (part of first identification unit)
22 second extraction unit (part of second identification unit)
24 Narrowing down judgment unit (part of first identification unit, part of second identification unit, first judgment unit)
26 Third extraction unit (part of second determination unit)
28 matching degree determination unit (a part of the second determination unit)
30 registered phrase determination unit (dictionary editing unit)

Claims (8)

A first for extracting a plurality of character strings having a common part from a first dictionary for registering character strings extracted from a log of a first device and specifying the types of the plurality of character strings extracted a specific part;
a second identification unit that extracts a character string including the common part from the log of the first device and identifies the type of the extracted character string;
a first determining unit that determines whether the type of character string specified by the first specifying unit is limited with respect to the type of character string specified by the second specifying unit;
If the result of the determination by the first determination unit is limited, a character string including the common part is extracted from the log of the second device different from the first device, and the extracted character a second determining unit that identifies the type of the string and determines whether the degree of matching between the identified character string type and the character string type identified by the second identifying unit is equal to or greater than a predetermined value;
a dictionary editing unit including the plurality of character strings extracted by the first identifying unit in a second dictionary for registering character strings extracted from the log of the second device when the degree of matching is equal to or greater than a predetermined value; and,
A dictionary creation device comprising: The first determining unit determines that the character string types specified by the second specifying unit include all of the character string types specified by the first specifying unit, and the second specifying unit specifies 2. The dictionary creation device according to claim 1, wherein when the type of the character string specified does not completely match the type of the character string specified by the first specifying unit, it is determined that the character string is limited. . If the result of the determination by the first determining unit is that the dictionary editing unit is not restrictive, the dictionary editing unit stores the information for extracting all character strings having the common part from the log of the second device. 3. The dictionary creation device according to claim 1, wherein the dictionary is included in the second dictionary. The dictionary editing unit includes information for extracting all character strings including the common part from the log of the second device in the second dictionary when the degree of matching is not equal to or greater than a predetermined value. A dictionary creation device according to any one of claims 1 to 3. The dictionary editing unit includes the character string including the common portion in the second dictionary when the second determination unit cannot extract the character string including the common portion from the log of the second device. The dictionary creation device according to any one of claims 1 to 4, characterized in that there is no 4. The dictionary editing unit includes, when a character string including a portion where an arbitrary character is specified is registered in the first dictionary, the character string to be included in the second dictionary. 6. The dictionary creation device according to any one of 1 to 5. 7. The dictionary creation according to any one of claims 1 to 6, characterized in that the process of identifying the type of the character string is a process of identifying the type of a portion of the character string other than a common portion. Device. extracting a plurality of character strings having a common part from a first dictionary for registering character strings extracted from the log of the first device, identifying the types of the plurality of extracted character strings;
Extracting a character string containing the common part from the log of the first device, identifying the type of the extracted character string,
determining whether the type of character string identified from the first dictionary is limited to the type of character string identified from the log of the first device;
As a result of the determination, if it is limited, a character string partly including the common part is extracted from the log of the second device different from the first device, and the type of the extracted character string is specified. and determining whether the degree of matching between the type of the specified character string and the type of the character string specified from the log of the first device is equal to or greater than a predetermined value;
If the degree of matching is equal to or greater than a predetermined value, include the plurality of character strings extracted from the first dictionary in a second dictionary for registering character strings extracted from the log of the second device;
A dictionary creation method characterized in that the processing is executed by a computer.

Images